ghc-hetmet.git
22 years ago[project @ 2001-05-16 09:55:27 by rrt]
rrt [Wed, 16 May 2001 09:55:27 +0000 (09:55 +0000)]
[project @ 2001-05-16 09:55:27 by rrt]
Add -mwin32 to CFLAGS for win32.

22 years ago[project @ 2001-05-16 09:46:43 by rrt]
rrt [Wed, 16 May 2001 09:46:43 +0000 (09:46 +0000)]
[project @ 2001-05-16 09:46:43 by rrt]
Move global options for utilities here.

22 years ago[project @ 2001-05-16 09:40:11 by sewardj]
sewardj [Wed, 16 May 2001 09:40:11 +0000 (09:40 +0000)]
[project @ 2001-05-16 09:40:11 by sewardj]
Add enough entries to RTS_MINGW_ONLY_SYMBOLS to make packages
{std lang concurrent util data text} work.  Also add a proper fix
to this using GetModuleHandle(NULL); unfortunately it doesn't
work for some unknown reason.

22 years ago[project @ 2001-05-15 19:57:08 by sof]
sof [Tue, 15 May 2001 19:57:08 +0000 (19:57 +0000)]
[project @ 2001-05-15 19:57:08 by sof]
-fglasgow-exts doesn't seem to imply -package data anymore -- add it to HC_OPTS

22 years ago[project @ 2001-05-15 19:54:46 by sof]
sof [Tue, 15 May 2001 19:54:46 +0000 (19:54 +0000)]
[project @ 2001-05-15 19:54:46 by sof]
Use LD_OPTS when linking the test binary

22 years ago[project @ 2001-05-15 15:29:03 by sewardj]
sewardj [Tue, 15 May 2001 15:29:03 +0000 (15:29 +0000)]
[project @ 2001-05-15 15:29:03 by sewardj]
Fix up the PEi386 linker well enough for ghci to start and function
on Win32.  Two potential problems still to be fixed properly:

* ocGetNames_PEi386 doesn't copy local syms into oc's local symtab,
  and it probably should (since the ELF one does).  Easy to fix.

* I can't figure out how to read syms in the executable itself
  on Win32.  This problem is solved properly for Unix-ELF by doing
  dlopen(NULL).  Here I have kludged it using RTS_MINGW_ONLY_SYMBOLS,
  but this is not a good long-term solution.

22 years ago[project @ 2001-05-15 15:01:48 by sewardj]
sewardj [Tue, 15 May 2001 15:01:48 +0000 (15:01 +0000)]
[project @ 2001-05-15 15:01:48 by sewardj]
Handle leading underscores correctly, for ghci on Win32.

22 years ago[project @ 2001-05-14 16:40:54 by sewardj]
sewardj [Mon, 14 May 2001 16:40:54 +0000 (16:40 +0000)]
[project @ 2001-05-14 16:40:54 by sewardj]
Record-selector unfolding fix #2

22 years ago[project @ 2001-05-14 12:55:31 by simonpj]
simonpj [Mon, 14 May 2001 12:55:31 +0000 (12:55 +0000)]
[project @ 2001-05-14 12:55:31 by simonpj]
comment wibble

22 years ago[project @ 2001-05-14 12:55:11 by simonpj]
simonpj [Mon, 14 May 2001 12:55:11 +0000 (12:55 +0000)]
[project @ 2001-05-14 12:55:11 by simonpj]
Add DEBUG code for CgInfo field of IdInfo

22 years ago[project @ 2001-05-14 12:03:41 by simonpj]
simonpj [Mon, 14 May 2001 12:03:41 +0000 (12:03 +0000)]
[project @ 2001-05-14 12:03:41 by simonpj]
**** MERGE WITH 5.00 BRANCH     ********

-----------------------------------------------
Expose record selectors with correct unfoldings
-----------------------------------------------

The problem was that there was that with "ghc --make" we were
passing on record selectors to importing modules in their un-tidied
form.  The idea was that they have very stylised bindings so they
don't need tidying.  But they do, because they may mention error
messages that get floated out.

Here's the comment from CoreTidy.lhs

-- MINOR-HACK ALERT: we *do* tidy record selectors.  Reason: they mention error
-- messages, which may be floated out:
-- x_field pt = case pt of
-- Rect x y -> y
-- Pol _ _  -> error "buggle wuggle"
-- The error message will be floated out so we'll get
-- lvl5 = error "buggle wuggle"
-- x_field pt = case pt of
-- Rect x y -> y
-- Pol _ _  -> lvl5
--
-- When this happens, it's vital that the Id exposed to importing modules
-- (by ghci) mentions lvl5 in its unfolding, not the un-tidied version.
--
-- What about the Id in the TyCon?  It probably shouldn't be in the TyCon at
-- all, but in any case it will have the error message inline so it won't matter.

22 years ago[project @ 2001-05-14 10:20:58 by sewardj]
sewardj [Mon, 14 May 2001 10:20:58 +0000 (10:20 +0000)]
[project @ 2001-05-14 10:20:58 by sewardj]
Change wording of panic message on encountering unboxed tuples to:

        Bytecode generator can't handle unboxed tuples.  Possibly due
        to foreign import/export decls in source.  Workaround:
        compile this module to a .o file, then restart session.

22 years ago[project @ 2001-05-10 22:55:27 by sof]
sof [Thu, 10 May 2001 22:55:27 +0000 (22:55 +0000)]
[project @ 2001-05-10 22:55:27 by sof]
Make lndir work with (current) cygwin impls + support the use of DOS-like paths (e.g, 'f:/oo')

22 years ago[project @ 2001-05-10 09:51:28 by rrt]
rrt [Thu, 10 May 2001 09:51:28 +0000 (09:51 +0000)]
[project @ 2001-05-10 09:51:28 by rrt]
Enhanced. Probably not to be used now, though.

22 years ago[project @ 2001-05-10 09:00:28 by simonmar]
simonmar [Thu, 10 May 2001 09:00:28 +0000 (09:00 +0000)]
[project @ 2001-05-10 09:00:28 by simonmar]
typo in error msg, pointed out by Alastair Reid.

22 years ago[project @ 2001-05-09 13:48:18 by simonmar]
simonmar [Wed, 9 May 2001 13:48:18 +0000 (13:48 +0000)]
[project @ 2001-05-09 13:48:18 by simonmar]
Add test for pattern parse errors, in particular getting the line
number right (4.08.2 didn't).

22 years ago[project @ 2001-05-09 13:46:29 by simonpj]
simonpj [Wed, 9 May 2001 13:46:29 +0000 (13:46 +0000)]
[project @ 2001-05-09 13:46:29 by simonpj]
Fix bug in spec-constr rule generation [Sergei2]

22 years ago[project @ 2001-05-09 13:28:11 by simonpj]
simonpj [Wed, 9 May 2001 13:28:11 +0000 (13:28 +0000)]
[project @ 2001-05-09 13:28:11 by simonpj]
**** MERGE WITH 5.00 BRANCH     ********

-------------------------------
Fix a rather obscure rule bogon
-------------------------------

The problem was that there was

class Foo a where
  op :: a -> a

{-# RULES "op" op x = x #-}

or something like that.  We attach locally defined rules, like this one,
to the local binding, in SimplCore.prepareRules.  Alas op doesn't reply
"True" to isLocalId, because it's a class selector (so it's a GlobalId
throughout).   Result: we treated the rule as an imported rule, and
therefore gave 'op' a fresh unique (becuase it looked as if it was
already in scope).  This only blew up in ghc --make or --interactive.

The handling of local vs global rules is rather unsatisfactory.
Something to muse on.

22 years ago[project @ 2001-05-09 13:05:07 by simonpj]
simonpj [Wed, 9 May 2001 13:05:07 +0000 (13:05 +0000)]
[project @ 2001-05-09 13:05:07 by simonpj]
Fix yesterdays bogons in parsing do-expressions; MERGE IN BRANCH

22 years ago[project @ 2001-05-09 13:04:30 by simonpj]
simonpj [Wed, 9 May 2001 13:04:30 +0000 (13:04 +0000)]
[project @ 2001-05-09 13:04:30 by simonpj]
Remove unused fn getContext

22 years ago[project @ 2001-05-09 09:55:41 by simonmar]
simonmar [Wed, 9 May 2001 09:55:41 +0000 (09:55 +0000)]
[project @ 2001-05-09 09:55:41 by simonmar]
Use -hcsuf to avoid clobbering .hc files when building several ways,
and clean up the -osuf usage (put all the suffix-changing options in
one place).

22 years ago[project @ 2001-05-09 09:43:44 by simonmar]
simonmar [Wed, 9 May 2001 09:43:44 +0000 (09:43 +0000)]
[project @ 2001-05-09 09:43:44 by simonmar]
Document -hcsuf option

22 years ago[project @ 2001-05-09 09:38:18 by simonmar]
simonmar [Wed, 9 May 2001 09:38:18 +0000 (09:38 +0000)]
[project @ 2001-05-09 09:38:18 by simonmar]
Add a new option: -hcsuf <suffix> which renames the .hc suffix to
<suffix> for this compilation, in the same way as -osuf and -hisuf.

To even things up, remove partial support for '-ohi -', which
allegedly dumped the .hi file to stdout but in reality never worked.
It's a strange thing to want to do anyway, but in any case you could
always say '-ohi /dev/stdout', or even 'ghc Foo.hs && cat Foo.hi'.

22 years ago[project @ 2001-05-08 17:33:57 by qrczak]
qrczak [Tue, 8 May 2001 17:33:57 +0000 (17:33 +0000)]
[project @ 2001-05-08 17:33:57 by qrczak]
Fix typo when times() is used.
Move foreign imports inside appropriate #ifdefs.

A small problem remained (besides Win32): For hpux getrusage is defined
here as a macro, and foreign importing a macro relies on compilation
via C. Should be wrapped in a function, but I'm not sure if it's safe
to call it getrusage.

22 years ago[project @ 2001-05-08 16:47:25 by sewardj]
sewardj [Tue, 8 May 2001 16:47:25 +0000 (16:47 +0000)]
[project @ 2001-05-08 16:47:25 by sewardj]
Insert interim fix in the bytecode gen to ignore polymorphic case
for the time being.  I can't see any way to fix it right in the
timescale before 5.00.1 goes out.  This works well enough to
make Sergei's DoCon thing run on the interpreter without segfaults.

   -- Nasty hack; treat
   --     case scrut::suspect of bndr { DEFAULT -> rhs }
   --     as
   --     let bndr = scrut in rhs
   --     when suspect is polymorphic or arrowtyped
   -- So the required strictness properties are not observed.
   -- At some point, must fix this properly.

22 years ago[project @ 2001-05-08 16:25:30 by simonpj]
simonpj [Tue, 8 May 2001 16:25:30 +0000 (16:25 +0000)]
[project @ 2001-05-08 16:25:30 by simonpj]
Fix braino

22 years ago[project @ 2001-05-08 14:54:58 by rrt]
rrt [Tue, 8 May 2001 14:54:58 +0000 (14:54 +0000)]
[project @ 2001-05-08 14:54:58 by rrt]
fixdll.pl renamed decyg.pl, for greater justice.
runexe.c added: it's a little Windows program which takes a command line and
runs it, without any connection to the parent process. This allows cygwin
binaries to be run against our patched cygwin1.dll (under a pseudonym),
without any connection being made if the parent process is also Cygwinised.

22 years ago[project @ 2001-05-08 14:44:37 by simonpj]
simonpj [Tue, 8 May 2001 14:44:38 +0000 (14:44 +0000)]
[project @ 2001-05-08 14:44:37 by simonpj]
**** MERGE WITH 5.00 BRANCH     ********

--------------------------------------
Make parallel list comprehensions work
--------------------------------------

There were two bugs

1.  The desugaring in DsListComp was generating code that failed Lint.
    I've restructured it quite a lot.

2.  More seriously, in a ParStmt, the last 'stmt' may be a guard;
    but previously both guards and the result of a list comprehension
    were encoded as an ExprStmt (see HsExpr.Stmt), using the fact that
    the stmt was last in the list to make the difference between a guard
    and a result.  But in parallel list comp this isn't right:

[ e | x <- xs, guard | y <- ys ]

    Here 'guard' is last in its list, but isn't an overall result.

    The sensible fix is to properly distinguish
"here's the answer"   (ResultStmt)
"here's a guard or an imperative action" (ExprStmt)

    The fix is rather easy, but touched quite a lot of files.  On the
    way I tidied up the parser a little.

22 years ago[project @ 2001-05-08 14:39:27 by simonpj]
simonpj [Tue, 8 May 2001 14:39:27 +0000 (14:39 +0000)]
[project @ 2001-05-08 14:39:27 by simonpj]
Add parallel list comp test

22 years ago[project @ 2001-05-08 14:33:01 by simonmar]
simonmar [Tue, 8 May 2001 14:33:01 +0000 (14:33 +0000)]
[project @ 2001-05-08 14:33:01 by simonmar]
oops, backout accidentally committed changes

22 years ago[project @ 2001-05-08 11:55:24 by simonmar]
simonmar [Tue, 8 May 2001 11:55:24 +0000 (11:55 +0000)]
[project @ 2001-05-08 11:55:24 by simonmar]
Fix setSrcLocP

22 years ago[project @ 2001-05-08 11:07:30 by simonmar]
simonmar [Tue, 8 May 2001 11:07:30 +0000 (11:07 +0000)]
[project @ 2001-05-08 11:07:30 by simonmar]
oops, remove debugging output

22 years ago[project @ 2001-05-08 10:58:48 by simonmar]
simonmar [Tue, 8 May 2001 10:58:48 +0000 (10:58 +0000)]
[project @ 2001-05-08 10:58:48 by simonmar]
Add

DriverPhases.haskellish_src_suffix :: String -> Bool
DriverPhases.haskellish_src_file   :: String -> Bool

which return True for suffixes of Haskell source files only.  The
existing haskellish_suffix and haskellish_file return True also for
.raw_s and .hc files.

We use these instead of haskellish_file in Main.main when deciding
whether to preprocess a file.

Fixes: compilation of .raw_s files, and potential bugs with
compilation of .hc files.

22 years ago[project @ 2001-05-08 08:55:17 by simonmar]
simonmar [Tue, 8 May 2001 08:55:18 +0000 (08:55 +0000)]
[project @ 2001-05-08 08:55:17 by simonmar]
Drop in CPUTime library replacement.  Still needs to be ported to
Win32.

22 years ago[project @ 2001-05-07 20:23:24 by sof]
sof [Mon, 7 May 2001 20:23:24 +0000 (20:23 +0000)]
[project @ 2001-05-07 20:23:24 by sof]
Extraneous -O

22 years ago[project @ 2001-05-07 16:19:36 by sof]
sof [Mon, 7 May 2001 16:19:36 +0000 (16:19 +0000)]
[project @ 2001-05-07 16:19:36 by sof]
couple of bugfixes to make the script do its thing

22 years ago[project @ 2001-05-07 16:18:49 by sof]
sof [Mon, 7 May 2001 16:18:49 +0000 (16:18 +0000)]
[project @ 2001-05-07 16:18:49 by sof]
support for building .pdf docs

22 years ago[project @ 2001-05-07 14:38:15 by simonmar]
simonmar [Mon, 7 May 2001 14:38:15 +0000 (14:38 +0000)]
[project @ 2001-05-07 14:38:15 by simonmar]
Give slightly more accurate line numbers for certain pattern parse errors.

22 years ago[project @ 2001-05-07 14:37:00 by simonmar]
simonmar [Mon, 7 May 2001 14:37:00 +0000 (14:37 +0000)]
[project @ 2001-05-07 14:37:00 by simonmar]
add
setSrcLocP :: SrcLoc -> P a -> P a

22 years ago[project @ 2001-05-07 11:42:31 by simonmar]
simonmar [Mon, 7 May 2001 11:42:31 +0000 (11:42 +0000)]
[project @ 2001-05-07 11:42:31 by simonmar]
typo in Word64 code pointed out by Alastair Reid.

22 years ago[project @ 2001-05-04 16:36:38 by simonmar]
simonmar [Fri, 4 May 2001 16:36:38 +0000 (16:36 +0000)]
[project @ 2001-05-04 16:36:38 by simonmar]
comment explaining why we check the perms on . before reading ./.ghci.

22 years ago[project @ 2001-05-04 15:18:00 by simonmar]
simonmar [Fri, 4 May 2001 15:18:00 +0000 (15:18 +0000)]
[project @ 2001-05-04 15:18:00 by simonmar]
.raw_s is a haskellish_suffix.

22 years ago[project @ 2001-05-04 15:12:16 by simonmar]
simonmar [Fri, 4 May 2001 15:12:16 +0000 (15:12 +0000)]
[project @ 2001-05-04 15:12:16 by simonmar]
document :set -package

22 years ago[project @ 2001-05-04 14:56:53 by simonmar]
simonmar [Fri, 4 May 2001 14:56:53 +0000 (14:56 +0000)]
[project @ 2001-05-04 14:56:53 by simonmar]
- only read ~/.ghci if it is owned by the current user and isn't
  writable by anyone else.

- Only read ./.ghci if both . and ./.ghci are owned by the current
  user and aren't writable by anyone else.  I think this is
  sufficient: we don't need to check .. and ../.. etc. because "."
  always refers to the same directory while a process is running.

- Don't load .so libraries in a package if that package is already
  linked with GHCi.  This stops us re-linking libm, libreadline etc.

- Allow packages to be loaded from within GHCi using
:set -package <name>
  NOTE: this will unload all modules currently loaded into the
  interpreter.  I did this to be on the safe side - I think perhaps
  it isn't necessary, but I haven't thought it through fully yet.

- fix CompManager.cmUnload in the process.  It was wrong in several
  ways.

MERGE WITH 5.00

22 years ago[project @ 2001-05-04 14:53:44 by simonpj]
simonpj [Fri, 4 May 2001 14:53:44 +0000 (14:53 +0000)]
[project @ 2001-05-04 14:53:44 by simonpj]
Add mutual recursion between types test

22 years ago[project @ 2001-05-04 14:44:07 by simonpj]
simonpj [Fri, 4 May 2001 14:44:07 +0000 (14:44 +0000)]
[project @ 2001-05-04 14:44:07 by simonpj]
Add more TC tracing

22 years ago[project @ 2001-05-04 14:43:49 by simonmar]
simonmar [Fri, 4 May 2001 14:43:49 +0000 (14:43 +0000)]
[project @ 2001-05-04 14:43:49 by simonmar]
touch# should be declared to have_side_effects.  This fixes the
remaining problem with safe foreign imports and ForeignPtrs.

22 years ago[project @ 2001-05-04 14:43:26 by simonpj]
simonpj [Fri, 4 May 2001 14:43:26 +0000 (14:43 +0000)]
[project @ 2001-05-04 14:43:26 by simonpj]
**** MERGE WITH 5.00 BRANCH     ********

--------------------------------
Fix a black hole when type checking type decls
--------------------------------

GHC was falling into a black hole when type checking a recursive
group of type declarations including a chain of type synonyms.

  type PhraseFun = PMap -> Float
  type PMap      = () -> Player
  data Player    = P.MkT P.PhraseFun

Reason: too much consistency checking in TcMonoType.
Easily fixed using the existing wimp_out hack, but it's a mess.
This commit fixes it for the 5.00 branch but I'll do something
better in the head shortly.

22 years ago[project @ 2001-05-04 14:40:42 by simonpj]
simonpj [Fri, 4 May 2001 14:40:42 +0000 (14:40 +0000)]
[project @ 2001-05-04 14:40:42 by simonpj]
Comments only

22 years ago[project @ 2001-05-04 14:23:52 by rrt]
rrt [Fri, 4 May 2001 14:23:52 +0000 (14:23 +0000)]
[project @ 2001-05-04 14:23:52 by rrt]
Script to change "cygwin1.dll" to something else, so that we can have our
own version of the cygwin DLL that doesn't clash with any installed DLLs,
for great justice.

22 years ago[project @ 2001-05-04 13:21:05 by simonmar]
simonmar [Fri, 4 May 2001 13:21:05 +0000 (13:21 +0000)]
[project @ 2001-05-04 13:21:05 by simonmar]
add
flushPackageCache :: [PackageConfig] -> IO ()

for modifying the list of available packages.

22 years ago[project @ 2001-05-04 09:37:13 by qrczak]
qrczak [Fri, 4 May 2001 09:37:13 +0000 (09:37 +0000)]
[project @ 2001-05-04 09:37:13 by qrczak]
Since overlapping rules (when one is more specific than the other)
seem to be handled well, add remaining fromIntegral rules to have
optimal conversions in all cases (given available primops).

22 years ago[project @ 2001-05-04 08:10:30 by simonpj]
simonpj [Fri, 4 May 2001 08:10:31 +0000 (08:10 +0000)]
[project @ 2001-05-04 08:10:30 by simonpj]
Comments only

22 years ago[project @ 2001-05-04 08:09:42 by simonpj]
simonpj [Fri, 4 May 2001 08:09:42 +0000 (08:09 +0000)]
[project @ 2001-05-04 08:09:42 by simonpj]
**** MERGE WITH 5.00 BRANCH     ********

--------------------------------
Make CPR work only for small products
--------------------------------

GHC was barfing when doing CPR for programs involving very large
products.  A one-line fix in MkId.mkDataCon makes it only do CPR for
smaller products (I chose 10).  Comments

-- We do not treat very big tuples as CPR-ish:
-- a) for a start we get into trouble because there aren't
--    "enough" unboxed tuple types (a tiresome restriction,
--    but hard to fix),
-- b) more importantly, big unboxed tuples get returned mainly
--    on the stack, and are often then allocated in the heap
--    by the caller.  So doing CPR for them may in fact make
--    things worse.

This should fix the error
Failed to find interface decl for
PrelGHC.(#,,,,,,,,,,,,,,,,,,,,,,,,
  ,,,,,,,,,,,,,,,,,,,,,,,,,,,
  ,,,,,,,,,,,,,,,,,,,,,,,#)

I hope.

22 years ago[project @ 2001-05-03 21:07:21 by sof]
sof [Thu, 3 May 2001 21:07:21 +0000 (21:07 +0000)]
[project @ 2001-05-03 21:07:21 by sof]
win32 tweak

22 years ago[project @ 2001-05-03 19:03:27 by qrczak]
qrczak [Thu, 3 May 2001 19:03:27 +0000 (19:03 +0000)]
[project @ 2001-05-03 19:03:27 by qrczak]
Use infix syntax on lhs of some rules.

22 years ago[project @ 2001-05-03 16:32:55 by simonmar]
simonmar [Thu, 3 May 2001 16:33:27 +0000 (16:33 +0000)]
[project @ 2001-05-03 16:32:55 by simonmar]
silence gcc 2.96 warning

22 years ago[project @ 2001-05-03 15:53:47 by simonmar]
simonmar [Thu, 3 May 2001 15:53:47 +0000 (15:53 +0000)]
[project @ 2001-05-03 15:53:47 by simonmar]
oops, fix braino

22 years ago[project @ 2001-05-03 15:50:10 by simonmar]
simonmar [Thu, 3 May 2001 15:50:10 +0000 (15:50 +0000)]
[project @ 2001-05-03 15:50:10 by simonmar]
redirect CPP output to /dev/null; gcc 2.96 seems to behave differently
that 2.95 when the -E and -MMD flags are given.

22 years ago[project @ 2001-05-03 14:31:58 by simonmar]
simonmar [Thu, 3 May 2001 14:31:58 +0000 (14:31 +0000)]
[project @ 2001-05-03 14:31:58 by simonmar]
Arrange to call touch# on each ForeignObj# argument after a 'safe'
foreign import call.

This turned out to be the easiest place to implement it: if we'd done
it in CorePrep or CoreToStg, it would have been awkward to generate
the simplest code.  At least doing it in the desugarer gives the
simplifier a crack at it later.

23 years ago[project @ 2001-05-03 12:33:50 by simonpj]
simonpj [Thu, 3 May 2001 12:33:50 +0000 (12:33 +0000)]
[project @ 2001-05-03 12:33:50 by simonpj]
**** MERGE WITH 5.00 BRANCH     ********

--------------------------------
Monomorphism restriction for implicit parameters
--------------------------------

This commit tidies up the way in which monomorphic bindings
are dealt with, incidentally fixing a bug to do with implicit
parameters.

The tradeoffs concerning monomorphism and implicit paramters are
now documented in TcSimplify.lhs, and all the strategic choices
are made there (rather than in TcBinds where they were before).

I've continued with choice (B) -- which Jeff first implemented --
because that's what Hugs does, lacking any other consensus.

23 years ago[project @ 2001-05-03 12:30:35 by simonpj]
simonpj [Thu, 3 May 2001 12:30:35 +0000 (12:30 +0000)]
[project @ 2001-05-03 12:30:35 by simonpj]
Add another IP test

23 years ago[project @ 2001-05-03 09:36:43 by simonmar]
simonmar [Thu, 3 May 2001 09:36:43 +0000 (09:36 +0000)]
[project @ 2001-05-03 09:36:43 by simonmar]
Oops, xrefs to varlistentries are unsupported apparently.  Xref the
section instead.

23 years ago[project @ 2001-05-03 09:35:28 by simonpj]
simonpj [Thu, 3 May 2001 09:35:28 +0000 (09:35 +0000)]
[project @ 2001-05-03 09:35:28 by simonpj]
Correct pretty print of HsIPVar

23 years ago[project @ 2001-05-03 09:35:09 by simonmar]
simonmar [Thu, 3 May 2001 09:35:09 +0000 (09:35 +0000)]
[project @ 2001-05-03 09:35:09 by simonmar]
mention, with a link, the +RTS -xc option.

23 years ago[project @ 2001-05-03 09:34:40 by simonmar]
simonmar [Thu, 3 May 2001 09:34:40 +0000 (09:34 +0000)]
[project @ 2001-05-03 09:34:40 by simonmar]
document the +RTS -xc option

23 years ago[project @ 2001-05-03 09:32:48 by simonpj]
simonpj [Thu, 3 May 2001 09:32:49 +0000 (09:32 +0000)]
[project @ 2001-05-03 09:32:48 by simonpj]
------------------------------------------------
Dramatically improve the error messages arising
from failed unifications triggered by 'improvement'
------------------------------------------------

A bit more plumbing in FunDeps, and consequential wibbles elsewhere

Changes this:

    Couldn't match `Int' against `[(String, Int)]'
Expected type: Int
Inferred type: [(String, Int)]

to this:

    Foo.hs:8:
Couldn't match `Int' against `[(String, Int)]'
    Expected type: Int
    Inferred type: [(String, Int)]
When using functional dependencies to combine
  ?env :: Int, arising from a type signature at Foo.hs:7
  ?env :: [(String, Int)],
    arising from use of implicit parameter `?env' at Foo.hs:8
When generalising the types for ident

23 years ago[project @ 2001-05-03 09:04:43 by simonpj]
simonpj [Thu, 3 May 2001 09:04:43 +0000 (09:04 +0000)]
[project @ 2001-05-03 09:04:43 by simonpj]
Improve error message when two signature tyvars get unified

23 years ago[project @ 2001-05-03 09:02:44 by simonpj]
simonpj [Thu, 3 May 2001 09:02:44 +0000 (09:02 +0000)]
[project @ 2001-05-03 09:02:44 by simonpj]
Error message wibble

23 years ago[project @ 2001-05-03 09:02:24 by simonpj]
simonpj [Thu, 3 May 2001 09:02:24 +0000 (09:02 +0000)]
[project @ 2001-05-03 09:02:24 by simonpj]
Tidy up output from -ddump-types

23 years ago[project @ 2001-05-03 09:01:50 by simonpj]
simonpj [Thu, 3 May 2001 09:01:50 +0000 (09:01 +0000)]
[project @ 2001-05-03 09:01:50 by simonpj]
Improve error message

23 years ago[project @ 2001-05-03 09:01:29 by simonpj]
simonpj [Thu, 3 May 2001 09:01:29 +0000 (09:01 +0000)]
[project @ 2001-05-03 09:01:29 by simonpj]
Wibble

23 years ago[project @ 2001-05-03 08:57:30 by simonmar]
simonmar [Thu, 3 May 2001 08:57:30 +0000 (08:57 +0000)]
[project @ 2001-05-03 08:57:30 by simonmar]
disable the hsc2hs suffix rules when booting from .hc files.

23 years ago[project @ 2001-05-03 08:52:52 by simonpj]
simonpj [Thu, 3 May 2001 08:52:52 +0000 (08:52 +0000)]
[project @ 2001-05-03 08:52:52 by simonpj]
Add a fundep fail test

23 years ago[project @ 2001-05-03 08:51:16 by simonpj]
simonpj [Thu, 3 May 2001 08:51:16 +0000 (08:51 +0000)]
[project @ 2001-05-03 08:51:16 by simonpj]
Remove bogus test

23 years ago[project @ 2001-05-03 08:47:06 by simonpj]
simonpj [Thu, 3 May 2001 08:47:06 +0000 (08:47 +0000)]
[project @ 2001-05-03 08:47:06 by simonpj]
Add implicit param test

23 years ago[project @ 2001-05-03 08:35:33 by simonpj]
simonpj [Thu, 3 May 2001 08:35:33 +0000 (08:35 +0000)]
[project @ 2001-05-03 08:35:33 by simonpj]
Add two more fundep tests

23 years ago[project @ 2001-05-03 08:18:35 by simonpj]
simonpj [Thu, 3 May 2001 08:18:35 +0000 (08:18 +0000)]
[project @ 2001-05-03 08:18:35 by simonpj]
Add a couple of tests

23 years ago[project @ 2001-05-03 08:13:39 by simonpj]
simonpj [Thu, 3 May 2001 08:13:39 +0000 (08:13 +0000)]
[project @ 2001-05-03 08:13:39 by simonpj]
Improve error message

23 years ago[project @ 2001-05-03 08:13:25 by simonpj]
simonpj [Thu, 3 May 2001 08:13:25 +0000 (08:13 +0000)]
[project @ 2001-05-03 08:13:25 by simonpj]
**** MERGE WITH 5.00 BRANCH     ********

--------------------------------
Fix a bad implicit parameter bug
--------------------------------

TcSimplify.tcSimplifyIPs was just completely wrong; it wasn't
doing improvement properly nor binding values properly. Sigh.

To make this work nicely I added
Inst.instName :: Inst -> Name

23 years ago[project @ 2001-05-03 08:10:09 by simonpj]
simonpj [Thu, 3 May 2001 08:10:09 +0000 (08:10 +0000)]
[project @ 2001-05-03 08:10:09 by simonpj]
Comment wibble only

23 years ago[project @ 2001-05-03 08:09:52 by simonpj]
simonpj [Thu, 3 May 2001 08:09:52 +0000 (08:09 +0000)]
[project @ 2001-05-03 08:09:52 by simonpj]
More to allow infix exprssions on LHS of rule

23 years ago[project @ 2001-05-03 08:08:44 by simonpj]
simonpj [Thu, 3 May 2001 08:08:44 +0000 (08:08 +0000)]
[project @ 2001-05-03 08:08:44 by simonpj]
Allow infix exprssions on LHS of rule

23 years ago[project @ 2001-05-03 08:08:12 by simonpj]
simonpj [Thu, 3 May 2001 08:08:12 +0000 (08:08 +0000)]
[project @ 2001-05-03 08:08:12 by simonpj]
Comment wibble

23 years ago[project @ 2001-05-02 18:19:27 by qrczak]
qrczak [Wed, 2 May 2001 18:19:27 +0000 (18:19 +0000)]
[project @ 2001-05-02 18:19:27 by qrczak]
Oops, exprIsConApp_maybe should ignore only InlineMe notes.

23 years ago[project @ 2001-05-02 16:15:09 by sewardj]
sewardj [Wed, 2 May 2001 16:15:09 +0000 (16:15 +0000)]
[project @ 2001-05-02 16:15:09 by sewardj]
Add test to check contexts on existential types in the interpreter.

23 years ago[project @ 2001-05-02 15:26:05 by rrt]
rrt [Wed, 2 May 2001 15:26:05 +0000 (15:26 +0000)]
[project @ 2001-05-02 15:26:05 by rrt]
Remove list of packages to install on Windows; the new story is that you
install all of Cygwin.

23 years ago[project @ 2001-05-02 12:14:49 by simonpj]
simonpj [Wed, 2 May 2001 12:14:49 +0000 (12:14 +0000)]
[project @ 2001-05-02 12:14:49 by simonpj]
A bit more about flags for generics

23 years ago[project @ 2001-05-01 16:01:06 by simonmar]
simonmar [Tue, 1 May 2001 16:01:06 +0000 (16:01 +0000)]
[project @ 2001-05-01 16:01:06 by simonmar]
Close files eagerly after analysing their imports.

23 years ago[project @ 2001-05-01 15:12:43 by rrt]
rrt [Tue, 1 May 2001 15:12:43 +0000 (15:12 +0000)]
[project @ 2001-05-01 15:12:43 by rrt]
Add zlib to list of required packages for building.

23 years ago[project @ 2001-05-01 13:11:16 by sewardj]
sewardj [Tue, 1 May 2001 13:11:16 +0000 (13:11 +0000)]
[project @ 2001-05-01 13:11:16 by sewardj]
Implement bci_STKCHECK.

23 years ago[project @ 2001-05-01 12:01:01 by simonmar]
simonmar [Tue, 1 May 2001 12:01:01 +0000 (12:01 +0000)]
[project @ 2001-05-01 12:01:01 by simonmar]
Not quite sure what this test does, but I've had it lying around for a while.

23 years ago[project @ 2001-05-01 11:51:50 by simonmar]
simonmar [Tue, 1 May 2001 11:51:50 +0000 (11:51 +0000)]
[project @ 2001-05-01 11:51:50 by simonmar]
Add an item about the networking library rewrite.

23 years ago[project @ 2001-05-01 11:46:56 by simonmar]
simonmar [Tue, 1 May 2001 11:46:56 +0000 (11:46 +0000)]
[project @ 2001-05-01 11:46:56 by simonmar]
Start a new directory for networking tests, and add a simple test
which starts two threads which communicate via a socket.

The socket subdirectory already contains some tests, but they're about
6 years old and don't appear to be designed to run without user
intervention.

23 years ago[project @ 2001-05-01 11:37:18 by sewardj]
sewardj [Tue, 1 May 2001 11:37:18 +0000 (11:37 +0000)]
[project @ 2001-05-01 11:37:18 by sewardj]
Ignore tyvars on variables bound by case alts when generating unpack code.

23 years ago[project @ 2001-05-01 11:06:19 by simonmar]
simonmar [Tue, 1 May 2001 11:06:19 +0000 (11:06 +0000)]
[project @ 2001-05-01 11:06:19 by simonmar]
Tests for BlockedOnDeadMVar and ThreadKilled exceptions (they generate
no output by default in forked threads).

23 years ago[project @ 2001-05-01 09:26:10 by simonmar]
simonmar [Tue, 1 May 2001 09:26:10 +0000 (09:26 +0000)]
[project @ 2001-05-01 09:26:10 by simonmar]
Don't include the win32 and com packages unless we're building for Windows.

23 years ago[project @ 2001-05-01 09:22:45 by simonmar]
simonmar [Tue, 1 May 2001 09:22:45 +0000 (09:22 +0000)]
[project @ 2001-05-01 09:22:45 by simonmar]
fix HC bootstrapping after LibsReadline no longer has the -l prefixes.

23 years ago[project @ 2001-05-01 09:16:55 by qrczak]
qrczak [Tue, 1 May 2001 09:16:56 +0000 (09:16 +0000)]
[project @ 2001-05-01 09:16:55 by qrczak]
Inline instance dictionary functions.
Remove {-# INLINE instance #-} support and uses.

23 years ago[project @ 2001-05-01 09:10:32 by simonmar]
simonmar [Tue, 1 May 2001 09:10:32 +0000 (09:10 +0000)]
[project @ 2001-05-01 09:10:32 by simonmar]
Add some {-# SCC #-} annotations, and fix a space leak.